A $2 million donation will launch new research at the Institute for Advanced Study (IAS) in Princeton to forge an understanding of how machine learning evolves.

Machine learning — sometimes called the leading edge of artificial intelligence — is the rapidly developing computer technology behind self-driving cars, complex web searches, medical and science applications, and face and speech recognition. Machine-learning programs synthesize knowledge in a way that's analogous to how children learn. The programs take examples, generalize, and then develop rules and understanding about the world without being taught directly. With time, the programs become better at particular tasks.

Avi Wigderson heads the computer science group at the Institute for Advanced Study’s school of mathematics.
handout
Avi Wigderson heads the computer science group at the Institute for Advanced Study’s school of mathematics.

"Machine-learning techniques are finding new applications almost daily and are already transforming society in numerous ways." said Avi Wigderson, who heads the institute's computer-science group at the School of Mathematics. "But because of the self-teaching aspect they have, their immense size, and the way they evolve, the final algorithms are not well understood by their own designers."

The gift, from Silicon Valley philanthropists Eric and Wendy Schmidt, will fund a three-year program focusing on the mathematical underpinnings of machine learning. Eric Schmidt is the executive chairman of Alphabet Inc., the parent company of Google. Wendy Schmidt is president of the Schmidt Family Foundation. IAS is best known as the academic home of some of the 20th century's greatest mathematical minds, including Albert Einstein, John von Neumann, and Kurt Godel.

Sanjeev Arora, of neighboring Princeton University, will lead the new program in theoretical machine learning as a visiting professor in the institute's School of Mathematics.